FMCW Radar Signal Processing

Overview

This work focuses on key radar signal processing techniques developed for Human Activity Recognition (HAR). The core components of the pipeline include:

• Range Profile Mapping: The initial step that extracts range information from reflected radar signals, revealing the location of targets.

• Range-Doppler Mapping: Simultaneously displays target velocity and range, essential for detecting and analyzing motion patterns.

• Micro-Doppler Spectrograms: Provides time-frequency analysis that highlights fine motion characteristics like arm swings, walking cycles, and other limb dynamics.

Preprocessing Techniques

These are applied after range profile mapping to enhance data quality and improve interpretability:

• Range Gating: Narrows down processing to a specific distance window, helping focus on human targets while suppressing clutter.

• DC Subtraction: Eliminates low-frequency biases from stationary background reflections.

• IQ Balancing: Corrects amplitude and phase mismatches in radar’s in-phase (I) and quadrature (Q) components for cleaner analysis.

• MTI Filtering (Moving Target Indication): Suppresses reflections from static objects, highlighting only the dynamic elements in the scene.

Visual Results

All stages of processing—including range profiles, range-Doppler maps, and micro-Doppler spectrograms—are visualized to demonstrate how each step improves signal clarity and supports activity classification.

Range Profile Mapping
1. Directly from 2D Raw Radar Data,
2. After Range Gating, DC Substraction and IQ balancing.
3. After doing MTI filtering

Range Doppler Mapping
1. Without Range Gating,
2. After Range Gating, DC Substraction and IQ balancing.
3. After doing MTI filtering

Micro-Doppler Spectrogram
1. Without Range Gating,
2. After Range Gating, DC Substraction and IQ balancing.
3. After doing MTI filtering